Proportion Priors for Image Sequence Segmentation

TitleProportion Priors for Image Sequence Segmentation
Publication TypeConference Paper
Year of Publication2013
AuthorsNieuwenhuis, C., Strekalovskiy E., & Cremers D.
Other Numbers3623

We propose a convex multilabel framework for imagesequence segmentation which allows to impose proportionpriors on object parts in order to preserve their size ratiosacross multiple images. The key idea is that for strongly deformable objects such as a gymnast the size ratio of respective regions (head versus torso, legs versus full body, etc.)is typically preserved. We propose different ways to imposesuch priors in a Bayesian framework for image segmentation. We show that near-optimal solutions can be computedusing convex relaxation techniques. Extensive qualitativeand quantitative evaluations demonstrate that the proportion priors allow for highly accurate segmentations, avoiding seeping-out of regions and preserving semantically relevant small-scale structures such as hands or feet. Theynaturally apply to multiple object instances such as players in sports scenes, and they can relate different objectsinstead of object parts, e.g. organs in medical imaging.The algorithm is efficient and easily parallelized leadingto proportion-consistent segmentations at runtimes aroundone second.


This work was partially funded by the Deutscher Akademischer Austausch Dienst (DAAD) through a postdoctoral fellowship.

Bibliographic Notes

Proceedings of the International Conference on Computer Vision 2013 (ICCV 2013), Sydney, Australia

Abbreviated Authors

C. Nieuwenhuis, E. Strekalovskiy, and D. Cremers

ICSI Research Group


ICSI Publication Type

Article in conference proceedings