11 publications
Work across 3D reconstruction and 3D generative models, and the representations underneath them — from LiDAR scenes to part-level mesh generation.
SIGGRAPH Asia 2026
Project leadJiahao Chang, Dong Du, Wanhu Sun, Yujian Zheng, Chuanyu Pan, Bowen Zhao, Chongjie Ye, Yuanming Hu, Xiaoguang Han
ACM SIGGRAPH Asia Conference Papers
Click a part on a rendered view of a mesh and get it back as a complete mesh — part extraction posed as conditional generation, not segmentation.
Preprint 2026
Project leadZhengyuan Li, Chuanyu Pan, Yuanming Hu, Raymond Yeh
arXiv preprint
Generates skeletons by tokenizing branches rather than BFS order, and fuses predictions across rotated views of the mesh at test time.
Preprint 2026
Project leadJeonghwan Kim, Yushi Lan, Yongwei Chen, Hieu Trung Nguyen, Chuanyu Pan, Xingang Pan
arXiv preprint
Builds a 3D scene from one image by splitting the work across specialist agents for initialization, environment construction and refinement.
CVPR 2026 · Oral presentation
Project leadYihao Luo, Xianglong He, Chuanyu Pan, Yiwen Chen, Jiaqi Wu, Yangguang Li, Wanli Ouyang, Yuanming Hu, Guang Yang, ChoonHwai Yap
IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
A sparse voxel representation that skips iso-surface extraction entirely, keeping sharp edges and internal structure at 2048³ resolution.
Preprint 2026
Chongjie Ye, Cheng Cao, Chuanyu Pan, Yiming Hao, Yihao Zhi, Yuanming Hu, Xiaoguang Han
arXiv preprint
Treats text, images and 3D as one token sequence, so plentiful 2D data acts as a geometric prior where 3D training data is scarce.
CVPR Workshop 2025 · Mobile AI Workshop
Zixun Huang, Keling Yao, Seth Z. Zhao, Chuanyu Pan, Allen Y. Yang
IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
A transformer for RGBD 6DoF pose that holds up under depth-sensor noise, with DTTD-Mobile, a dataset captured on an iPhone 14 Pro.
CoRL 2024
Chuanyu Pan, Aolin Xu
Conference on Robot Learning (CoRL)
Turns unlabelled LiDAR into a dense 3D opacity grid via differentiable volume rendering, then forecasts future point clouds from it.
CGI 2023 · Journal of Virtual Reality & Intelligent Hardware (VRIH)
Chuanyu Pan, Guowei Yang, Taijiang Mu, Yu-Kun Lai
Computer Graphics International (CGI)
Turns a single portrait photo into an animatable 3D cartoon face, combining StyleGAN stylization with a two-stage textured reconstruction.
CVPR Workshop 2023 · 2nd Workshop Challenge on Vision Datasets Understanding
Weiyu Feng*, Seth Z. Zhao*, Chuanyu Pan*, Adam Chang, Yichen Chen, Zekun Wang, Allen Y. Yang
IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · * equal contribution
An RGB-D dataset for longer-range 3D object tracking: 103 scenes of 10 objects, with per-pixel masks and motion-capture ground truth.
ICLR 2022
Chuanyu Pan*, Yanchao Yang*, Kaichun Mo, Yueqi Duan, Leonidas J. Guibas
International Conference on Learning Representations (ICLR) · * equal contribution
Learns object-centric representations continually, using a hypernetwork to synthesize discriminative per-object weights from latent codes.
CVPR 2020 · Oral presentation
Zhe Li, Tao Yu, Chuanyu Pan, Zerong Zheng, Yebin Liu
IEEE Conference on Computer Vision and Pattern Recognition (CVPR)
Detailed 3D self-portraits in seconds from one RGBD camera, pairing learning-based recovery with volumetric non-rigid fusion.