-
Graph Representation Learning
Graph Neural Networks (GNNs) and Graph Convolutional Networks (GCNs) generalize the power of CNN to learn on graph-structured data (e.g., citation networks, molecules/proteins, 3D meshes). However, existing GNNs cannot be trained deeply and has a limited application range. Our research line includes both theoretical analysis of GNNs and exploration for their potential applications:
-
Neural Representation and Optimization
An emerging perspective towards neural networks is to treat them as universal approximators of arbitrary continuous functions. One can solve many inverse problems by: 1) representing the objective function as a multi-layer perceptron, 2) writing a differentiable forward process, and 3) adopting gradient-based algorithms to reach the optimum. In particular, we are using neural representation framework to solve inter-disciplinary challenges in physical imaging, medical imaging, and biological microscopy:
-
3D Human Avatar Modeling
Human avatars are one of the most important elements in virtual reality. Our research line includes 3D human body reconstruction, 3D human clothing reconstruction/simulation, 3D pose estimation, human motion capture, and action understanding:
-
Towards Competitive Diving Video Understanding
Submitted to ECCV 2020
Competitive diving video understanding by identifying the dive number, assessing the performance, 2D pose/action recovery.
[Paper]
[Video]
[Revision]
-
3D Clothing Segmentation via Graph Convolutional Networks
![]()
Garment segmentation networks based on graph convolutional networks for human models.
[Slides]
[Code]
[Dataset]
-
Towards 3D Human Shape Recovery Under Clothing
Submitted to ICCV 2019
A learning-based scheme for robustly and accurately estimating clothing fitness as well as the human shape on clothed 3D human scans.
[Paper]
[Video]
[Dataset]
[Revision]
Keep thy heart with all diligence; for out of it are the issues of life.
I was born and raised up in Chengdu, southwestern China, together with my parents and my brother.
I learned C/C++ programming languages at 12 years old, and spent my leisure time developing myBase Desktop 7 during high school.
I am now in Shanghai, eastern China to pursue my Bachelor's degree at ShanghaiTech University with a focus on machine learning and computer vision. I enjoy doing research, reading insightful papers, and writing computer programs. I believe good work should deliver either mathematical motivations or extensive engineering efforts.
I am also a core team memember of GeekPie Association, with whom I played Hackthathon and took part in lots of innovation and development design-sprint events.
