Junzhe Lu

News

• 2026.09Awarded the 2026 National Scholarship for Graduate Students.
• 2026.07ArtiAtlas accepted to SIGGRAPH Asia 2026.
• 2026.04Seed3D 2.0 released by ByteDance Seed.
• 2026.01ViMoGen and EchoMotion accepted to ICLR 2026.
• 2025.10Our paper on consistency models was published in IEEE TAI.
• 2025.08Started as a 3D Foundation Model Algorithm Intern at ByteDance Seed.
• 2025.07NOVA3D selected as an ICME 2025 Best Paper Candidate (15/3737).
• 2025.06DPoser-X accepted to ICCV 2025 as an Oral (64 of 11,239 submissions; 0.57%).

Publications

ArtiAtlas: A Holistic Ecosystem for Diverse 3D Articulation Modeling

Junzhe Lu, Jing Lin, Ziang Cao, Haoqian Wang, Ziwei Liu, Jianfeng Zhang

SIGGRAPH Asia 2026To Appear

Builds the data curation and annotation pipeline, develops ArtiPipe for automatic joint annotation, and designs and trains Artisan to recover articulated structures from static meshes.

The Quest for Generalizable Motion Generation: Data, Model, and Evaluation

Jing Lin*, Ruisi Wang*, Junzhe Lu*, Ziqi Huang, Guorui Song, Ailing Zeng, Xian Liu, Chen Wei, Wanqi Yin, Qingping Sun, Zhongang Cai, Lei Yang, Ziwei Liu

ICLR 2026

Introduces the ViMoGen-228K dataset, a motion generator combining motion-capture and video-generation priors, and the MBench evaluation benchmark.

Project Code arXiv

Toward a Unified Framework for Consistency Generative Modeling

Hongkun Dou, Junzhe Lu, Jinyang Du, Chengwei Fu, Wen Yao, Hongjue Li, Yue Deng

IEEE TAI 2026

Paper

NOVA3D: Normal Aligned Video Diffusion Model for Single Image to 3D Generation

Yuxiao Yang, Peihao Li, Yuhong Zhang, Junzhe Lu, Xianglong He, Minghan Qin, Weitao Wang, Haoqian Wang

ICME 2025Best Paper Candidate · 15/3737

arXiv

Score-Based Neural Processes

Hongkun Dou, Junzhe Lu, Zeyu Li, Xiaoqing Zhong, Wen Yao, Lijun Yang, Yue Deng

IEEE TNNLS 2025

Paper

SEGSID: A Semantic-Guided Framework for Sonar Image Despeckling

Shaohua Liu, Junzhe Lu, Hongkun Dou, Jiajun Li, Yue Deng

IEEE TIP 2025

Paper

ScoreSeg: Leveraging Score-Based Generative Model for Self-Supervised Semantic Segmentation of Remote Sensing

Junzhe Lu, Guangjun He, Hongkun Dou, Qing Gao, Leyuan Fang, Yue Deng

IEEE JSTARS 2023

Code Paper

Seed3D 2.0: Advancing High-Fidelity Simulation-Ready 3D Content Generation

Seed 3D Team — Junzhe Lu (Contributor)

Tech Report 2026ByteDance Seed

Contributed data filtering and augmentation operators to the foundation-model data pipeline.

Project Paper

Experience

Education

Honors & Awards