On Learning to Localize Objects With Minimal Supervision

TitleOn Learning to Localize Objects With Minimal Supervision
Publication TypeConference Paper
Year of Publication2014
AuthorsSong, H. Oh, Girshick R., Jegelka S., Mairal J., Harchaoui Z., & Darrell T.
Other Numbers3685
Abstract

Learning to localize objects with minimal supervision is an important problem in computer vision, since large fully annotated datasets are extremely costly to obtain. In this paper, we propose a new method that achieves this goal withonly image-level labels of whether the objectsare present or not. Our approach combines a discriminative submodular cover problem for automatically discovering a set of positive object windows with a smoothed latent SVM formulation.The latter allows us to leverage efficient quasi-Newton optimization techniques. Our experiments demonstrate that the proposed approach

URLhttps://www.icsi.berkeley.edu/pubs/vision/learninglocalize14.pdf
Bibliographic Notes

Proceedings of the 31st International Conference in Machine Learning (ICML), Beijing, China

Abbreviated Authors

H. O. Song, R. Girshick, S. Jegelka, J. Mairal, Z. Harchaoui, and T. Darrell

ICSI Research Group

Vision

ICSI Publication Type

Article in conference proceedings