Crowd4D: Scene-Aware Monocular 4D Crowd Reconstruction
Reconstructs scene-consistent 4D crowds from monocular video by jointly optimizing human motion and large-scale scene geometry.
About Me
I am Hongbo Kang (康洪菠), a Ph.D. candidate at Tianjin University advised by Prof. Kun Li. I also collaborate closely with Prof. Yu-Kun Lai at Cardiff University. My research focuses on human-centered 3D reconstruction and generation, with an emphasis on perceiving, modeling, and simulating humans in complex scenes. I work on individual and crowd reconstruction, virtual mixed-age crowd data construction, and individual and crowd motion simulation, among other topics.
News
- My paper “AnnyCrowd: Mixed-Age Crowd Reconstruction from a Single Image” has been accepted to CICAI!
- I attended ICML 2026 and presented our work “Crowd4D”.
- I am honored to be awarded the China Scholarship Council (CSC) scholarship!
- My paper “Crowd4D: Scene-Aware Monocular 4D Crowd Reconstruction” has been accepted to ICML!
- Our DyCrowd poster will be presented at China3DV!
- Our paper “MuRE: Multi-Relationship Encoder for 3D Human Pose Estimation” has been accepted to CVIU!
- Our paper “DRPose: A Diffusion-based Pose Refinement Framework for 3D Human Pose Estimation” has been accepted to TCSVT!
- I am honored to be selected for CAST Youth Science and Technology Talent Cultivation Project Doctoral Student Special Program!
- Our paper “DBMambaPose: Decoupled Spatial-Temporal Bidirectional State Space Model for Efficient 3D Human Pose Estimation” has been accepted to PR!
- My paper “DyCrowd: Towards Dynamic Crowd Reconstruction from a Large-scene Video” has been accepted to TPAMI!
- Our paper “RESCUE: Crowd Evacuation Simulation via Controlling SDM-United Characters” has been accepted to ICCV as Highlight!
- My paper “Double-chain Graph Convolution Transformer for 3D Human Pose Estimation” has been accepted to TMM!
- I started my Ph.D. in Prof. Kun Li’s team at Tianjin University.
Selected Research
DyCrowd: Towards Dynamic Crowd Reconstruction from a Large-scene Video
Introduces the first framework to recover temporally consistent 3D poses, positions, and body shapes for hundreds of people from large-scene monocular video.
RESCUE: Crowd Evacuation Simulation via Controlling SDM-United Characters
Simulates real-time, personalized 3D crowd evacuation with human-inspired sensing, decision, and motion control.
Global and local spatio-temporal encoder for 3D human pose estimation
Combines global and local spatial-temporal encoding to capture full-body and joint-level dependencies in 3D pose estimation.
Publications
2026
AnnyCrowd: Mixed-Age Crowd Reconstruction from a Single Image
CAAI International Conference on Artificial Intelligence, 2026 · Paper · Project Page · Code
Crowd4D: Scene-Aware Monocular 4D Crowd Reconstruction
International Conference on Machine Learning, 2026 · Paper · Project Page · Code
MuRE: Multi-Relationship Encoder for 3D Human Pose Estimation
Computer Vision and Image Understanding, 2026 · Paper
DRPose: A Diffusion-based Pose Refinement Framework for 3D Human Pose Estimation
IEEE Transactions on Circuits and Systems for Video Technology, 2026 · Code
DBMambaPose: Decoupled Spatial-Temporal Bidirectional State Space Model for Efficient 3D Human Pose Estimation
Pattern Recognition, 2026 · Code
2025
DyCrowd: Towards Dynamic Crowd Reconstruction from a Large-scene Video
IEEE Transactions on Pattern Analysis and Machine Intelligence, 2025 · Paper · Project Page · Code
RESCUE: Crowd Evacuation Simulation via Controlling SDM-United Characters
International Conference on Computer Vision, 2025 · Paper · Project Page · Code
Double-chain Graph Convolution Transformer for 3D Human Pose Estimation
ICFNet: Interactive-complementary fusion network for monocular 3D human pose estimation
2024
Diffusion-based Pose Refinement and Multi-Hypothesis Generation for 3D Human Pose Estimation
International Conference on Acoustics, Speech, and Signal Processing, 2024 · Paper · Code
Hierarchical flow learning for low-light image enhancement
Digital Communications and Networks, 2024 · Paper
2023
Global and local spatio-temporal encoder for 3D human pose estimation
IEEE Transactions on Multimedia, 2023 · Paper
Awards
- China Scholarship Council (CSC) Scholarship
- CAST Youth Science and Technology Talent Cultivation Project Doctoral Student Special Program
- Outstanding Graduate Student of Chongqing (Top 1%)
- National Scholarship (Top 1%)
Academic Service
- Journal Reviewer: IJCV, TMM, TCSVT, PR, etc.
- Conference Reviewer: CVPR, AAAI, MM, SIGGRAPH Asia, etc.