3 years ago

Adversarially Optimizing Intersection over Union for Object Localization Tasks.

Brian D. Ziebart, Sima Behpour, Kris M. Kitani

An implicit uncertainty exists in the annotations of computer visions datasets due to annotator disagreement and the high-dimensional space that annotations must be selected from.Rather than attempting to remove all annotation uncertainty,which we view as hopeless, or ignoring it, which can be detrimental, we choose to embrace uncertainty in the design of our learning approach. Specifically, we address uncertainty adversarially by approximating provided datasets annotations within a game-theoretic formulation of prediction tasks. The adversarial approximator is constrained to resemble the training data annotations according to a set of specified features.This induces a learned feature-based potential function that we then apply to new test cases. We demonstrate the efficiency and predictive performance of our approach on the ILSVRC2012 image dataset, showing significant improvements over existing methods.

Publisher URL: http://arxiv.org/abs/1710.07735

DOI: arXiv:1710.07735v1

You might also like
Never Miss Important Research

Researcher is an app designed by academics, for academics. Create a personalised feed in two minutes.
Choose from over 15,000 academics journals covering ten research areas then let Researcher deliver you papers tailored to your interests each day.

  • Download from Google Play
  • Download from App Store
  • Download from AppInChina

Researcher displays publicly available abstracts and doesn’t host any full article content. If the content is open access, we will direct clicks from the abstracts to the publisher website and display the PDF copy on our platform. Clicks to view the full text will be directed to the publisher website, where only users with subscriptions or access through their institution are able to view the full article.