We are Urban Spatial Intelligence (USI) Research Group at the State Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing (LIESMARS), Wuhan University. We focus on 3D Computer Vision, particularly including 3D reconstruction, scene understanding, and point cloud processing as well as their applications in intelligent transportation system, digital twin cities, urban sustainable development, and robotics. Check our works by topic:
Our Team (click to expand):
- Lab Leaders
- Bisheng Yang(杨必胜): Professor and the head of LIESMARS, Wuhan University.
- Zhen Dong (董震): Professor and the head of 3S (GNSS/RS/GIS) integration department in the LIESMARS, Wuhan University.
- Chi Chen (陈驰): Associate professor at LIESMARS, Wuhan University.
- Academic Advisors
- Yuan Liu (刘缘): Incoming Assistant Professor at HKUST, working on 3D AIGC including neural rendering, neural representations, and 3D generative models.
- Xiaoxin Mi (米晓新): PostDoc at Wuhan University of Technology, working on scene understanding and modeling and intelligent transportation system.
- Activate Members
- Yuhao Li (李雨昊): PhD student in LIESMARS. Research interests include Mobile Laser Scanning Point Cloud, LiDAR SLAM, Multi-modality Fusion, Place Recognition, Retrieval and Localization.
- Xianghong Zou (邹响红): PhD student in LIESMARS. Research interest lies in the field of 3D Computer Vision, particularly including point cloud localization and 3D change detection.
- Haiping Wang (王海平): Ph.D. student at LIESMARS, interested in 3D reconstruction, 3D understanding, and 3D LLM.
- Chen Long (龙宸) PHD student at LIESMARS, interested in point cloud enhancement, 3D shape restoration, urban sustainable development.
Public datasets (click to expand):
- 📂 WHU-TLS
: TLS PC registration benchmark covering 11 scenarios;
- 📂 WHU-Helmet
: A helmet-based multi-sensor SLAM benchmark;
- 📂 WHU-Urban-3D : ALS/MLS semantic/instance segmentation benchmark;
- 📂 WHU-Railway3D
: Semantic segmentation benchmark for railway scenario;
- 📂 WHU-Lane
: A Benchmark Approach and Dataset for Large-scale Lane Mapping from MLS Point Clouds;
Point Cloud Registration (click to expand):
- 📂 BSC (ISPRS J'17)
: A handcrafted point cloud local descriptor utilizing CPU;
- 📂 YOHO (ACM MM'22)
: A learning-based point cloud local rotation-equivariant descriptor;
- 📂 RoReg (TPAMI'23)
: Utilizing rotation-equivariance in the whole pipeline of pairwise registration;
- 📂 SGHR (CVPR'23)
: A simple multiview pc registration baseline;
- 📂 MSReg (IEEE TGRS'24)
: Fast 4DOF registration of MLS and stereo point clouds;
Image-to-point cloud Registration (click to expand):
- 📂 FreeReg (ICLR'24)
: FreeReg extracts cross-modality features from pretrained diffusion models and monocular depth estimators for accurate zero-shot image-to-point cloud registration;
- 📂 CoFiI2P (RA-L'24)
: CoFiI2P is a coarse-to-fine framework for image-to-point cloud registration task;
Point Cloud Upsampling (click to expand):
- 📂 PC2-PU (ACM MM'22)
: A transformer-based point cloud upsampling baseline;
Point Cloud / Depth Completion (click to expand):
- 📂 KT-Net (AAAI'23)
: A transformer-based point cloud completion baseline;
- 📂 SparseDC (Information Fusion'24)
: Depth Completion from sparse and non-uniform inputs;
- 📂 EGIInet (ECCV'24)
: Single view image guided point cloud completion framework;
Point Cloud Localization (click to expand):
- 📂 PatchAugNet (ISPRS J'23)
: A cross-platform pc localization baseline;
- 📂 LAWS (ISPRS J'24)
: Regard point cloud localization as a classification problem;
Normal Estimation (click to expand):
- 📂 AdaFit (ICCV'21)
: Rethinking pc normal estimation;
Object Detection (click to expand):
- 📂 ME-Net (JAG'23)
: Objection detection utilizing both image and Lidar from mobile platform;
Image / Point Cloud Semantic Segmentation (click to expand):
- 📂 Mobile-Seed (RAL'24)
: An online framework for simultaneous semantic segmentation and boundary detection on compact robots;
Urban Morphology & Sustainable Development (click to expand):
- 📂 3DBIE-SolarPV (Applied Energy‘24)
: City-scale solar PV potential estimation on 3D buildings using multi-source RS data: A case study in Wuhan, China;
HDMap (click to expand):
- 📂 LaneMapping
