About me
I am a PhD student in Institute of Computing Technology, Chinese Academy of Sciences (ICT, CAS), advised by Prof. Zhaoqi Wang. I got my bachelor’s degree from Harbin Institute of Technology (HIT) in 2019.
My research interests lie in the field of computer vision and computer graphics, especially multi-view stereo, 3D reconstruction (large-scale scenes & human body) and 3D detection.
🔥 News
- 2022-07-21 Research Intern at VIS, Baidu Inc, first day working on 3D detection😋, supervised by Dr. Xin Wang & Prof. Jingdong Wang.
- 2022-04-25 One paper accepted by CGI2022💥
- 2022-04-13 Research Intern at IDL, Baidu Research, supervised by Tianyi Wu & Prof. Guodong Guo.
- 2019-09-01 Join VR Lab at Institute of Computing Technology, Chinese Academy of Sciences ICT, CAS, start my CV journey!🔫
- 2019-02-01 Research Asistant at VR Lab, Institute of Computing Technology, Chinese Academy of Sciences.
📝 Publications
![sym](images/CFDS3D.png)
CFDS3D: Centroid Feature Diffusion Sampling framework for 3D Object Detection
Haotian Hu, Fanyi Wang, Jingwen Su, Shiyu Gao, Yaonong Wang, Laifeng Hu, Yuan Zhou, Yuhan Wang, Jineng Han, Mingyang Zhang, Weiye Fang
- We introduce a point-based 3D detection framework, named CFDS3D, which achieves efficient and accurate detection in large-scale point clouds scenes when inserting to existing point-based methods
![sym](images/mscvpmvsnet.png)
Cost Volume Pyramid Network with Multi-strategies Range Searching
Shiyu Gao, Zhaoxin Li, Zhaoqi Wang
Project Link 👉: https://github.com/SibylGao/MSCVP-MVSNet
- We discuss the impact of different search strategies in each stage for pyramid-based MVS methods and propose MS-CVPMVSNet, a pyramid-based method which leverage multi-dimensional information.
- To further exploit information in cost volume of deep MVS, we propose unimodal assumption as a training label in second stage.
- Our method obtains SOTA results (especially in completeness) on DTU and BlendedMVS dataset.
💻 Project
Here are some selected projects in recent years if you are interested:
- 3D Human Animation Rendering Based on Unreal Engine: project link
- End-to-end Deep Learning Network for Multi-view Stereo: project link
- Traffic flow simulation & visualisation in Large Scale Evacuation Simulation System: project link
- Infantry Robot Vision System Design for ROBOMaster Robotics Competition: project link