Image-based 3D object detection for autonomous driving

PhD Qualifying Examination


Title: "Image-based 3D object detection for autonomous driving"

by

Mr. Qing LIAN


Abstract:

Image-based 3D object detection aims at identifying and localizing the 
surrounding obstacles in the 3D space, which plays an essential role in 
autonomous driving. In this survey, I will provide an overview of recent 
image-based 3D object detection methods in autonomous driving scenarios. I will 
first introduce the task definition and the related datasets in 3D object 
detection. Then I will present various top-performing methods, including 
direct-regression based, 2D-3d constraint based, pseudo-lidar and BEV based 
approaches. Next, I will introduce our empirical study on the direct-regression 
based methods and present the work that integrates the direct-regression and 
geometry-constraint based approaches. Finally, I will present several promising 
future research directions for camera-based 3D object detection, including 
temporal modeling, semi-supervised training, uncertainty quantification and 
etc.


Date:  			Thursday, 29 September 2022

Time:                  	2:00pm - 4:00pm

Venue:			Room 4475
 			lifts 25-26

Zoom meeting:		https://hkust.zoom.us/j/3178158301

Committee Members:	Prof. Tong Zhang (Supervisor)
 			Prof. Xiaofang Zhou (Chairperson)
 			Dr. Dan Xu
 			Dr. Yingcong Chen (AI Thrust)


**** ALL are Welcome ****