eduzhai > Applied Sciences > Engineering >

Vehicle Re-ID for Surround-view Camera System

  • Save

... pages left unread,continue reading

Document pages: 8 pages

Abstract: The vehicle re-identification (ReID) plays a critical role in the perceptionsystem of autonomous driving, which attracts more and more attention in recentyears. However, to our best knowledge, there is no existing complete solutionfor the surround-view system mounted on the vehicle. In this paper, we arguetwo main challenges in above scenario: i) In single camera view, it isdifficult to recognize the same vehicle from the past image frames due to thefisheye distortion, occlusion, truncation, etc. ii) In multi-camera view, theappearance of the same vehicle varies greatly from different camera sviewpoints. Thus, we present an integral vehicle Re-ID solution to addressthese problems. Specifically, we propose a novel quality evaluation mechanismto balance the effect of tracking box s drift and target s consistency.Besides, we take advantage of the Re-ID network based on attention mechanism,then combined with a spatial constraint strategy to further boost theperformance between different cameras. The experiments demonstrate that oursolution achieves state-of-the-art accuracy while being real-time in practice.Besides, we will release the code and annotated fisheye dataset for the benefitof community.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×