DMCA
Occlusion Boundary Detection and Figure/Ground Assignment from Optical Flow
Cached
Download Links
| Citations: | 27 - 2 self |
BibTeX
@MISC{Sundberg_occlusionboundary,
author = {Patrik Sundberg and Thomas Brox and Michael Maire and Pablo Arbeláez and Jitendra Malik},
title = {Occlusion Boundary Detection and Figure/Ground Assignment from Optical Flow},
year = {}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
In this work, we propose a contour and region detector for video data that exploits motion cues and distinguishes occlusion boundaries from internal boundaries based on optical flow. This detector outperforms the state-of-the-art on the benchmark of Stein and Hebert [24], improving average precision from.58 to.72. Moreover, the optical flow on and near occlusion boundaries allows us to assign a depth ordering to the adjacent regions. To evaluate performance on this edge-based figure/ground labeling task, we introduce a new video dataset that we believe will support further research in the field by allowing quantitative comparison of computational models for occlusion boundary detection, depth ordering and segmentation in video sequences. 1.
Keyphrases
optical flow occlusion boundary detection figure ground assignment motion cue video sequence video data depth ordering distinguishes occlusion boundary adjacent region new video dataset quantitative comparison internal boundary region detector edge-based figure ground computational model average precision occlusion boundary
