News
- [2024/08] I will serve as a reviewer for ICLR 2025.
- [2024/05] I will serve as a reviewer for NeurIPS 2024.
- [2024/04] I will attend VALSE 2024, see you in Chongqing!
- [2023/10] This website is established!
|
Research
My current research focuses on Diffusion for Perception, specifically, tapping the potential of visual-language knowledge contained in pre-trained diffusion models in discriminative tasks. For the task-wise perspective, I mainly focus on open-world scene understanding and relation perception.
Meanwhile, I am broadly interested in Computer Vision, Multimodal Learning, and Machine Learning topics.
|
|
Improving Diffusion-based Data Augmentation with Inversion Spherical Interpolation
Yanghao Wang,
Long Chen
arxiv, 2024
paper
We propose a novel Diffusion-based Inversion Interpolation DA method called Diff-II, which can generate faithful and diverse samples to expand the original training set in data-scarce scenarios.
|
|
Random Boxes Are Open-world Object Detectors
Yanghao Wang,
Zhongqi Yue, Xian-Sheng Hua,
Hanwang Zhang
ICCV, 2023
paper
/
code
We show that classifiers trained with random region proposals achieve state-of-the-art Open-world Object Detection with our designed matching score.
|
|
CyberRel: Joint entity and relation extraction for cybersecurity concepts
Yongyan Guo,
Zhengyu Liu,
Cheng Huang,
Jiayong Liu,
Wangyuan Jing,
Ziwang Wang,
Yanghao Wang
ICICS, 2021, (Best Student Paper)
paper
We proposed CyberRel, a joint entity and relation extraction model for cyber-security concepts.
|
Awards & Honors
- Excellent Graduate of Sichuan University (Bachelor's degree)
- Best Student Paper Award of ICICS 2021
- The First Prize in the 14th National Information Security Work Competition (2021)
|
|