Weakly-Supervised Discovery of Visual Pattern Configurations
Abstract
The prominence of weakly labeled data gives rise to a growing demand for object detection methods that can cope with minimal supervision. We propose an approach that automatically identifies discriminative configurations of visual patterns that are characteristic of a given object class. We formulate the problem as a constrained submodular optimization problem and demonstrate the benefits of the discovered configurations in remedying mislocalizations and finding informative positive and negative training examples. Together, these lead to state-of-the-art weakly-supervised detection results on the challenging PASCAL VOC dataset.
Cite
Text
Song et al. "Weakly-Supervised Discovery of Visual Pattern Configurations." Neural Information Processing Systems, 2014.Markdown
[Song et al. "Weakly-Supervised Discovery of Visual Pattern Configurations." Neural Information Processing Systems, 2014.](https://mlanthology.org/neurips/2014/song2014neurips-weaklysupervised/)BibTeX
@inproceedings{song2014neurips-weaklysupervised,
title = {{Weakly-Supervised Discovery of Visual Pattern Configurations}},
author = {Song, Hyun Oh and Lee, Yong Jae and Jegelka, Stefanie and Darrell, Trevor},
booktitle = {Neural Information Processing Systems},
year = {2014},
pages = {1637-1645},
url = {https://mlanthology.org/neurips/2014/song2014neurips-weaklysupervised/}
}