Occlusion-Aware Object Localization, Segmentation and Pose Estimation

Samarth Brahmbhatt, Heni Ben Amor and Henrik Christensen

Abstract

We present a learning approach for localization and segmentation of objects in an image in a manner that is robust to partial occlusion. Our algorithm produces a bounding box around the full extent of the object and labels pixels in the interior that belong to the object. Like existing segmentation aware detection approaches, we learn an appearance model of the object and consider regions that do not fit this model as potential occlusions. However, in addition to the established use of pairwise potentials for encouraging local consistency, we use higher order potentials which capture information at the level of image segments. We also propose an efficient loss function that targets both localization and segmentation performance. Our algorithm achieves 13.52% segmentation error and 0.81 area under the false-positive per image vs. recall curve on average over the challenging CMU Kitchen Occlusion Dataset. This is 42.44% less segmentation error and a 16.13% increase in localization performance compared to the state-of-the-art. Finally, we show that the visibility labelling produced by our algorithm can make full 3D pose estimation from a single image robust to occlusion.

Session

Poster 1

Files

Extended Abstract (PDF, 1235K)
Paper (PDF, 6M)
Supplemental Materials (ZIP, 4M)

DOI

10.5244/C.29.80
https://dx.doi.org/10.5244/C.29.80

Citation

Bibtex

@inproceedings{BMVC2015_80,
	title={Occlusion-Aware Object Localization, Segmentation and Pose Estimation},
	author={Samarth Brahmbhatt and Heni Ben Amor and Henrik Christensen},
	year={2015},
	month={September},
	pages={80.1-80.13},
	articleno={80},
	numpages={13},
	booktitle={Proceedings of the British Machine Vision Conference (BMVC)},
	publisher={BMVA Press},
	editor={Xianghua Xie, Mark W. Jones, and Gary K. L. Tam},
	doi={10.5244/C.29.80},
	isbn={1-901725-53-7},
	url={https://dx.doi.org/10.5244/C.29.80}
}