Hi there, this is Xingrui! 👋
Before JHU, I obtained my Master’s degree from University of Southern Califonia, working closely with Prof. Laurent Itti. Prior to USC, I recieved my B.S. of Statistics from Renmin University of China, supervised by Prof. Hanfang Yang. I also spent my time in the AI lab of Samsung R&D Institute, supervised by Dr. Yang Liu, and Wangxuan institute of computer technology (WICT), Peking University, supervised by Prof. Yongtao Wang.
My research interest mainly lie in computer vision, especially about the topic of scene understanding, interpretability, the human-centric vision system and robustness. I am also interested in the interactions between CV and NLP & Robotics (like visual question answering, embodied AI, language grounding)
Long term interest
Building computer vision system which: (1) can understand scenes effective (includes object position and hierarchical relationship); (2) has the ability of imageination and physical inference; (3) can be grounded to other modality (language, action, etc.)
My current research experience & work center around:
- Visual question answering. The out-of-domain generalization problem of VQA model amd the ability to understanding more complex and hierarchical 3D scenes in VQA task.
- Humanoid AI. Build computer vision system with more humanoid cognitive learning ability ( few shot recognization and imagination), by learning color, shape, texture separately.
- Reinforcement learning agents with human guidence.
[Sep 2023] One paper accepted by NeurIPS 2023.
[Mar 2023] The SuperCLEVR paper is selected as a highlight in CVPR 2023 (2.5% of submissions).
[Feb 2023] One paper accepted at CVPR 2023.
[Jul 2022] One paper accepted at ECCV 2022.