Zhenyang Li (李镇洋)

Zhenyang Li (李镇洋)

Ph.D. Student, HKU ECE

Language 语言

About关于我

I study how to reconstruct, generate, render, and simulate dynamic 3D/4D worlds, with current interests in world models, 3D Gaussian Splatting, and multimodal 3D generation.

我致力于动态 3D/4D 世界的重建、生成、渲染与仿真，目前重点关注世界模型、3D Gaussian Splatting 与多模态三维生成。

🔬 Research Focus🔬 研究方向

My work spans four connected directions, from core 3D/4D reconstruction and generation to perception, embodied intelligence, and cross-disciplinary applications.

我的研究涵盖四个相互关联的方向，从核心的 3D/4D 重建与生成延伸至场景感知、具身智能及交叉领域应用。

3D/4D Scene Reconstruction, Depth & Correspondence

Event-to-image reconstruction, NeRF, 3D Gaussian Splatting, stereo and multi-view depth estimation, and feature correspondence for recovering appearance, geometry, and motion.

Representative works: ERF-GS · E2Pano · 4DGS

Structured Generative Models, Neural Network-driven Rendering & Physical Simulation

Structured scene and garment generation, path-traced image synthesis, event-camera simulation, and video-based physical system identification for controllable, simulation-ready worlds.

Representative works: PatternGSL · Scenix · EventTracer

Video Understanding, Scene Perception & Embodied AI

Action recognition, anomaly detection, panoramic segmentation, vision-language-action reasoning, and embodied navigation for understanding and interacting with dynamic environments.

Representative works: O-VAD · SAP · ConsistNav

Computational Imaging, Biomedical AI & Engineering Systems

Neural holography, foundation models for biomedical imaging, AR-assisted healthcare, recommender systems, and wireless communication.

Representative works: 3D-HoloNet · CryoSAM · AR-assisted surgery

3D/4D 场景重建、深度估计与图像对应

研究事件到图像重建、NeRF、3D Gaussian Splatting、立体与多视图深度估计及特征对应，用于恢复场景外观、几何与运动。

代表性工作：ERF-GS · E2Pano · 4DGS

结构化生成模型、神经网络驱动的渲染与物理仿真

研究结构化场景与三维服装生成、路径追踪图像合成、事件相机仿真，以及基于视频的物理系统辨识，构建可控且可仿真的三维世界。

代表性工作：PatternGSL · Scenix · EventTracer

视频理解、场景感知与具身智能

研究动作识别、异常检测、全景分割、视觉-语言-动作推理与具身导航，使智能体理解动态环境并与之交互。

代表性工作：O-VAD · SAP · ConsistNav

计算成像、生物医学智能与工程系统

探索神经全息、生物医学影像基础模型、增强现实辅助医疗、推荐系统与无线通信。

代表性工作：3D-HoloNet · CryoSAM · 增强现实辅助手术

🔥 News🔥 新闻动态

  • 2026.08:  📄 Scenix is released on arXiv: sparse-view 3D scene reconstruction via executable scene programs.
  • 2026.08:  📄 E2Pano is released on arXiv: event-to-panorama image reconstruction.
  • 2026.07:  🏆 O-VAD is accepted to ECCV 2026: object-centric industrial video anomaly detection.
  • 2026.06:  🏆 ERF-GS is accepted to Computational Visual Media (CVMJ): fast-motion reconstruction from disjoint Event-RGB viewpoints.
  • 2026.05:  🏆 EventTracer is accepted to IEEE TVCG: fast path tracing-based event-stream rendering.
  • 2026.05:  📄 ConsistNav is released on arXiv: semantic executive control for zero-shot object navigation.
Show earlier news
  • 2026.04:  🏆 PatternGSL is accepted to SIGGRAPH 2026: structured, simulation-ready 3D garment modeling.
  • 2026.04:  📄 Our work on AR-assisted dental implant surgery is published in Journal of the Society for Information Display (JSID).
  • 2026.03:  🏆 Our work on stereo matching generalization is accepted to 3DV 2026.
  • 2026.03:  📄 SAP is released on arXiv: segmenting high-resolution 4K panoramas.
  • 2026.08:  📄 Scenix 已发布于 arXiv：通过可执行场景程序实现稀疏视角三维场景重建。
  • 2026.08:  📄 E2Pano 已发布于 arXiv：从事件流重建全景图像。
  • 2026.07:  🏆 O-VAD 被 ECCV 2026 接收：基于目标中心跟踪与推理的工业视频异常检测。
  • 2026.06:  🏆 ERF-GS 被 Computational Visual Media (CVMJ) 接收：从非共视事件与 RGB 视角重建快速运动。
  • 2026.05:  🏆 EventTracer 被 IEEE TVCG 接收：基于快速路径追踪的事件流渲染。
  • 2026.05:  📄 ConsistNav 已发布于 arXiv：通过语义执行控制改进零样本目标导航。
展开更早动态
  • 2026.04:  🏆 PatternGSL 被 SIGGRAPH 2026 接收：面向仿真的结构化三维服装建模。
  • 2026.04:  📄 增强现实辅助牙种植导航工作发表于 Journal of the Society for Information Display（JSID）。
  • 2026.03:  🏆 立体匹配泛化工作被 3DV 2026 接收。
  • 2026.03:  📄 SAP 已发布于 arXiv：面向高分辨率 4K 全景图像的实例分割。

📝 Publications📝 论文发表

🎯 Selected Publications (* indicates equal contribution)🎯 代表性论文（* 表示共同一作）

2026

2025

2024

2023

2022

🎖 Honors and Awards🎖 荣誉与奖励

🏆 Competitions

  • 2023.12 Champion 🥇, Guangdong-Hong Kong-Macao Greater Bay Area (Huangpu) International Algorithm Case Competition
    Topic: Three-dimensional reconstruction of objects using neural implicit representation

  • 2020.04 Top 7/256 🥈, High-energy particle collision classification challenge, Beindata

  • 2019.05 Second Prize 🥉, American College Student Mathematical Modeling Contest (MCM)

💰 Scholarships & Academic Honors

  • 2017 TE Connectivity Scholarship (Top 1/600), Beijing SMC Education Foundation Outstanding Scholarship Special Award (Top 1/600), Second Prize in National Mathematics Competition for College Students

🏆 竞赛获奖

  • 2023.12 冠军 🥇，粤港澳大湾区（黄埔）国际算法算例大赛
    赛题：基于神经隐式表示的物体三维重建

  • 2020.04 前 7 / 256 🥈，高能粒子碰撞分类挑战赛，Beindata

  • 2019.05 二等奖 🥉，美国大学生数学建模竞赛（MCM）

💰 奖学金与学术荣誉

  • 2017 TE Connectivity Scholarship（Top 1/600）、北京晟铭创科教育基金会优秀学生奖学金特等奖（Top 1/600）、全国大学生数学竞赛二等奖

📖 Education📖 教育经历

  • 🎓 电机与计算机工程博士（原电子电气工程） | 2023.09 - 至今
    香港大学（HKU）
    导师：Yifan (Evan) Peng & Jia Pan @ HKU

  • 🎓 大数据工程硕士 | 2020.09 - 2023.07
    清华大学（THU）
    自动化系、深圳国际研究生院
    导师：Kai Zhang 教授

  • 🎓 电子信息工程学士 | 2016.09 - 2020.07
    南京理工大学（NJUST）
    电子与光学工程学院

💻 Internships💻 实习经历

  • 🏢 Research Intern | Oct. 2025 - Present
    Tencent LIGHTSPEED STUDIOS, Shenzhen, Guangdong, China
    Research Focus: Multimodal & 3D Content Generation and Simulation

  • 🏢 Research Intern | Jul. 2022 - Nov. 2022
    Megvii Technology Limited (Face++), Beijing, China
    Research Focus: Visual Odometry, NeRF, Multi-View Stereo (MVS), Feature Matching

  • 🏢 Research Intern | Nov. 2021 - May. 2022
    Microsoft Research Asia (MSRA), Beijing, China
    Research Focus: Video Understanding, Learning-based Computer Vision

  • 🏢 Artificial Intelligence Researcher | Mar. 2021 - Sep. 2021
    Huawei Technologies Co., Ltd., Shenzhen, Guangdong, China
    Research Focus: 3D Reconstruction, Visual Localization

  • 🏢 研究实习生 | 2025.10 - 至今
    Tencent LIGHTSPEED STUDIOS，中国广东深圳
    研究方向：多模态与 3D 内容生成及仿真

  • 🏢 研究实习生 | 2022.07 - 2022.11
    Megvii Technology Limited (Face++)，中国北京
    研究方向：视觉里程计、NeRF、多视图立体（MVS）、特征匹配

  • 🏢 研究实习生 | 2021.11 - 2022.05
    Microsoft Research Asia (MSRA)，中国北京
    研究方向：视频理解、学习式计算机视觉

  • 🏢 人工智能研究员 | 2021.03 - 2021.09
    Huawei Technologies Co., Ltd.，中国广东深圳
    研究方向：三维重建、视觉定位

🧑‍⚖️ Academic Service🧑‍⚖️ 学术服务

Program Committee Member

  • AAAI 2027
  • 34th ACM Multimedia (ACMMM) 2026

Conference Reviewer

  • Vision & Graphics: CVPR 2026; ICCV 2025; ECCV 2026; SIGGRAPH Asia 2025 (XR Track); 3DV 2027, 2026, 2025; ISMAR 2026, 2025
  • AI & Machine Learning: NeurIPS 2026, 2025 (Main + Datasets and Benchmarks), 2024; ICML 2025; ICLR 2025; AISTATS 2025; ACML Conference 2026, 2025; ACML Journal Track 2026
  • Multimedia & Vision: ACM MM 2025; BMVC 2026

Journal Reviewer

  • IEEE Journal of Selected Topics in Signal Processing (J-STSP)

程序委员会成员

  • AAAI 2027
  • 第 34 届 ACM Multimedia（ACMMM）2026

会议审稿

  • 视觉与图形：CVPR 2026；ICCV 2025；ECCV 2026；SIGGRAPH Asia 2025（XR Track）；3DV 2027、2026、2025；ISMAR 2026、2025
  • 人工智能与机器学习：NeurIPS 2026、2025（Main + Datasets and Benchmarks）、2024；ICML 2025；ICLR 2025；AISTATS 2025；ACML Conference 2026、2025；ACML Journal Track 2026
  • 多媒体与视觉：ACM MM 2025；BMVC 2026

期刊审稿

  • IEEE Journal of Selected Topics in Signal Processing（J-STSP）

💬 Invited Talks & Teaching💬 报告与教学

  • 🎤 Workshop Organizer & Speaker | Dec. 2025
    WeLight Workshop, The University of Hong Kong
    Organized by WeLightLab@HKU

  • 🎤 Guest Lecturer | May 2023
    ELEC4544: AI and Deep Learning, The University of Hong Kong
    Invited by Dr. Yifan (Evan) Peng

  • 🎤 Workshop 组织者与报告人 | 2025.12
    WeLight Workshop，香港大学
    由 WeLightLab@HKU 组织

  • 🎤 特邀讲师 | 2023.05
    ELEC4544: AI and Deep Learning，香港大学
    由 Yifan（Evan）Peng 博士邀请