Ph.D. student wins Best Paper Award at WACV

 

January 15, 2015

ECE Ph.D. student Zhiding Yu received the Best Paper Award at the IEEE Winter Conference on Applications of Computer Vision (WACV 2015), one of the computer vision top conferences. The paper, “Structured Hough Voting for Vision-based Highway Border Detection,” was co-authored by ECE Professor Vijayakumar Bhagavatula and General Motors Company researchers Dr. Wende Zhang and Dr. Dan Levi.

The paper addresses the problem of automatically detecting highway borders and shoulders using computer vision and machine learning methods. The considered problem is of high significance for future autonomous driving systems since knowing the road structure is indispensable for subsequent decision making operations. It uses a strategy where a number of trained road border and lane marking detectors are triggered, followed by a novel model called “Structured Hough Voting” to generate corresponding detection of road border and shoulder. The proposed Structured Hough Voting model exploits both the temporal correlation of detections as well as the geometric relationship between road borders and lane markings. It also combines the diverse voting results from multiple different methods, therefore working robustly and accurately under various challenging scenarios.

Carnegie Mellon University is conducting broad collaborative research with the General Motors Company to develop future commercial autonomous driving vehicles. The winning paper is part of this collaborative project.