Object Detection and Segmentation from Joint Embedding of Parts and Pixels

TitleObject Detection and Segmentation from Joint Embedding of Parts and Pixels
Publication TypeConference Paper
Year of Publication2011
AuthorsMaire, M., Yu S. X., & Perona P.
Published inProceedings of International Conference on Computer Vision
Date Published09/2011
Keywordsangular embedding, figure-ground organization, object segmentation, poselets
Abstract

We present a new framework in which image segmentation, figure/ground organization, and object detection all appear as the result of solving a single grouping problem. This framework serves as a perceptual organization stage that integrates information from low-level image cues with that of high-level part detectors. Pixels and parts each appear as nodes in a graph whose edges encode both affinity and ordering relationships. We derive a generalized eigen-problem from this graph and read off an interpretation of the image from the solution eigenvectors. Combining an off-the-shelf top-down part-based person detector with our low-level cues and grouping formulation, we demonstrate improvements to object detection and segmentation.

Acknowledgment

ONR MURI N00014-06-1-0734, ONR MURI 1015 G NA127, and ARL Cooperative Agreement
W911NF-10-2-0016 supported this work. Stella X. Yu was funded
by NSF CAREER IIS-0644204 and a Clare Boothe Luce Profes-
sorship. Thanks to Jitendra Malik for suggesting poselets as a fig-
ure/ground cue and Lubomir Bourdev for providing poselet code.
 

URLhttp://www1.icsi.berkeley.edu/~stellayu/publication/doc/2011objICCV.pdf
ICSI Research Group

Vision