Oct 19, 2023 Thursday
Abstract:
In the realm of computer vision and video analysis, the quest for efficient yet effective methodologies has given rise to transformative research. In this talk, we will introduce innovative approaches that enhance computational efficiency without sacrificing performance. Firstly, we will introduce a novel mask propagation framework for efficient video semantic segmentation by propagating masks efficiently across video frames. This approach significantly reduces computational costs while achieving state-of-the-art performance. Then, we will introduce an efficient spatio-temporal pyramid transformer for action detection. This framework optimally balances accuracy and efficiency by incorporating local and global attention mechanisms, resulting in a model that outperforms existing methods with reduced computational complexity. The confluence of these works underscores the vital importance of efficiency in video understanding, with applications spanning autonomous vehicles, surveillance, and entertainment.
Time:
Oct 19, 2023 Thursday
16:30-17:20
Location:
RmE1-101, GZ Campus
Zoom:
628 334 1826 (PW: 234567)
Speaker Bio:
![](https://ait.hkust-gz.edu.cn/wp-content/uploads/2023/10/image-5.png)
Dr. Xiaojun Chang
Full Professor, Australian Artificial Intelligence Institute, University of Technology Sydney
Dr. Xiaojun Chang is a Full Professor at the Faculty of Engineering and Information Technology Sydney, Australia. He is the Director of the ReLER Lab at the Australian Artificial Intelligence Institute (AAII) and a recipient of the Australian Research Council (ARC) Discovery Early Career Researcher Award (DECRA). His primary research areas include multimodal learning, computer vision, and green artificial intelligence, with a focus on their applications in social goods. He has led more than ten national-level projects, including the ARC Research Grants and the Linkage Program in Australia. His research findings have been published in over 150 international top-tier journals and CCF A-level conferences, including T-PAMI and TIP. His papers have been cited over 13,000 times on Google Scholar. He has been recognized as a highly cited researcher by Clarivate Analytics from 2019 to 2023. Additionally, he serves as an associate editor for prestigious international journals like IEEE TCSVT and ACM TOMM and acts as an Area Chair for CCF A-level conferences.