Profile image
Shuaiyi Huang 黄帅一

About Me

I am a 1st-year Ph.D. student in Computer Science at University of Maryland, College Park, working with Prof. Abhinav Shrivastava. Before that, I obtained my Master's degree in Computer Science from ShanghaiTech University, China in 2020, under the supervision of Prof. Xuming He. And I got my Bachelor's degree in Software Engineering at Tongji University, China in 2017.

My research interests broadly include Computer Vision and Deep Learning, with a focus on scene understanding and low-level vision using strong or weak supervision. I am especially interested in correspondence, segmentation, occlusion-aware scene understanding, visual question answering, and video analysis.


  • 11/2020: One paper is submitted to CVPR 2021
  • 07/2020: One paper is accepted to PRCV 2020
  • 10/2019: I attended ICCV 2019 at Soul, Korea
  • 07/2019: One paper is accepted to ICCV 2019
  • 10/2017: I attended ICCV 2017 at Venice, Italy
  • 07/2017: One paper is accepted to ICCV 2017


Ph.D. in Department of Computer Science, University of Maryland, College Park, USA
Sep. 2020 - present
M.Sc. in School of Information Science and Technology, ShanghaiTech University, Shanghai, China
Sep. 2017 - present
B.E. in School of Software Engineering, Tongji University, Shanghai, China
Sep. 2013 - Jun. 2017

Publications & Projects

Confidence-aware Adversarial Learning for Self-supervised Semantic Matching
Shuaiyi Huang, Qiuyue Wang, Xuming He
Chinese Conference on Pattern Recognition and Computer Vision (PRCV), 2020
Semantic Correspondence Self supervision Uncertainty Estimation Generative model
Dynamic Context Correspondence Network for Semantic Alignment
Shuaiyi Huang, Qiuyue Wang, Songyang Zhang, Shipeng Yan, Xuming He
IEEE International Conference on Computer Vision (ICCV), 2019
Semantic Correspondence Weak supervision
Amodal Instance Segmentation
Shuaiyi Huang, Xuming He
We aim at predicting the complete mask of an object containing both its visible and occluded parts. Key insights: (i) introducing human priors on complete shape is important given limited training data; (ii) developing a robust representation for occlusion is critical considering the drawback of holistic object representations. Three key parts in our pipeline: (i) a reference set as a memory bank of complete masks; (ii) a novel visual concept based voting network for part-based similarity learning and instance center alignment; (iii) a shape transfer network based on graph neural network for final mask prediction.
Research Project, 2018
Segmentation Part-based representations for object center voting Graph neural networks
Structured Attentions for Visual Question Answering
Chen Zhu, Yanpeng Zhao, Shuaiyi Huang, Kewei Tu, Yi Ma
IEEE International Conference on Computer Vision (ICCV), 2017
Visual quesiton answering Conditional Random Field