AI Thrust Seminar| Towards Robust and Holistic Scene Understanding for Autonomous Driving and Helping People with Visual Impairments

3:15pm - 4:15pm
Meeting ID:998 7898 5575 Passcode:AIT123

Both autonomous driving systems and mobility assistance systems for visually impaired people need to sense the surrounding information to facilitate upper-level navigational applications. We observe that robust and holistic scene understanding are sharing challenges for perception in self-driving systems and mobility assistance systems. This talk will first discuss the sharing challenges in both perception systems and question whether vision transformers are ready for real-world scene segmentation in surrounding understanding. Then, this talk will introduce multiple research works for robust scene understanding covering RGB-X semantic segmentation, transparent object segmentation, and accident scene segmentation. Further, this talk will further introduce multiple research works for holistic scene understanding covering panoramic semantic segmentation, panoramic panoptic segmentation, and LiDAR data top-view segmentation. Finally, this talk will introduce our established Vision4Blind system for supporting navigation assistance for people with visual impairments.

講者/ 表演者:
Dr. Kailun Yang
Karlsruhe Institute of Technology (KIT)

Dr. Kailun Yang is a postdoctoral researcher at the Computer Vision for Human-Computer Interaction (CV:LAB) Lab at Karlsruhe Institute of Technology (KIT), working with Prof. Rainer Stiefelhagen. Prior to that, he received his Ph.D. degree from Zhejiang University (ZJU), working with Prof. Kaiwei Wang and Prof. Jian Bai. He conducted a Ph.D. internship at the RobeSafe research group at the University of Alcalá (UAH), working with Prof. Luis M. Bergasa. His research focuses on real-time computer vision, multi-modal optical sensing, panoramic scene understanding, and their applications for intelligent vehicles and navigation assistance systems for people with visual impairments. His research works published in diverse areas spanning computer vision, optics, robotics, human-computer interaction, and intelligent transportation systems have received more than 1600 citations. He is the recipient of a Best Paper Award at the 2021 IEEE Intelligent Vehicles Symposium (IV).

語言
英文
適合對象
校友
教職員
公眾
科大家庭
研究生
本科生
主辦單位
Artificial Intelligence Thrust, HKUST(GZ)
新增活動
請各校內團體將活動發布至大學活動日曆。