English

Image Co-localization by Mimicking a Good Detector's Confidence Score Distribution

Computer Vision and Pattern Recognition 2016-07-26 v2

Abstract

Given a set of images containing objects from the same category, the task of image co-localization is to identify and localize each instance. This paper shows that this problem can be solved by a simple but intriguing idea, that is, a common object detector can be learnt by making its detection confidence scores distributed like those of a strongly supervised detector. More specifically, we observe that given a set of object proposals extracted from an image that contains the object of interest, an accurate strongly supervised object detector should give high scores to only a small minority of proposals, and low scores to most of them. Thus, we devise an entropy-based objective function to enforce the above property when learning the common object detector. Once the detector is learnt, we resort to a segmentation approach to refine the localization. We show that despite its simplicity, our approach outperforms state-of-the-art methods.

Keywords

Cite

@article{arxiv.1603.04619,
  title  = {Image Co-localization by Mimicking a Good Detector's Confidence Score Distribution},
  author = {Yao Li and Linqiao Liu and Chunhua Shen and Anton van den Hengel},
  journal= {arXiv preprint arXiv:1603.04619},
  year   = {2016}
}

Comments

Accepted to Proc. European Conf. Computer Vision 2016