Advanced

Reinforcement learning for visual object detection

Mathe, Stefan; Pirinen, Aleksis LU and Sminchisescu, Cristian LU (2016) 2016 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2016 In 2016 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2016 2016-January. p.2894-2902
Abstract

One of the most widely used strategies for visual object detection is based on exhaustive spatial hypothesis search. While methods like sliding windows have been successful and effective for many years, they are still brute-force, independent of the image content and the visual category being searched. In this paper we present principled sequential models that accumulate evidence collected at a small set of image locations in order to detect visual objects effectively. By formulating sequential search as reinforcement learning of the search policy (including the stopping condition), our fully trainable model can explicitly balance for each class, specifically, the conflicting goals of exploration - sampling more image regions for better... (More)

One of the most widely used strategies for visual object detection is based on exhaustive spatial hypothesis search. While methods like sliding windows have been successful and effective for many years, they are still brute-force, independent of the image content and the visual category being searched. In this paper we present principled sequential models that accumulate evidence collected at a small set of image locations in order to detect visual objects effectively. By formulating sequential search as reinforcement learning of the search policy (including the stopping condition), our fully trainable model can explicitly balance for each class, specifically, the conflicting goals of exploration - sampling more image regions for better accuracy-, and exploitation - stopping the search efficiently when sufficiently confident about the target's location. The methodology is general and applicable to any detector response function. We report encouraging results in the PASCAL VOC 2012 object detection test set showing that the proposed methodology achieves almost two orders of magnitude speed-up over sliding window methods.

(Less)
Please use this url to cite or link to this publication:
author
organization
publishing date
type
Chapter in Book/Report/Conference proceeding
publication status
published
subject
in
2016 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2016
volume
2016-January
pages
9 pages
publisher
IEEE Computer Society
conference name
2016 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2016
external identifiers
  • scopus:84986295261
ISBN
9781467388511
DOI
10.1109/CVPR.2016.316
language
English
LU publication?
yes
id
7a9e0d3d-4779-4f7e-8633-b43597a6338f
date added to LUP
2017-02-08 13:54:44
date last changed
2017-11-12 04:29:12
@inproceedings{7a9e0d3d-4779-4f7e-8633-b43597a6338f,
  abstract     = {<p>One of the most widely used strategies for visual object detection is based on exhaustive spatial hypothesis search. While methods like sliding windows have been successful and effective for many years, they are still brute-force, independent of the image content and the visual category being searched. In this paper we present principled sequential models that accumulate evidence collected at a small set of image locations in order to detect visual objects effectively. By formulating sequential search as reinforcement learning of the search policy (including the stopping condition), our fully trainable model can explicitly balance for each class, specifically, the conflicting goals of exploration - sampling more image regions for better accuracy-, and exploitation - stopping the search efficiently when sufficiently confident about the target's location. The methodology is general and applicable to any detector response function. We report encouraging results in the PASCAL VOC 2012 object detection test set showing that the proposed methodology achieves almost two orders of magnitude speed-up over sliding window methods.</p>},
  author       = {Mathe, Stefan and Pirinen, Aleksis and Sminchisescu, Cristian},
  booktitle    = {2016 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2016},
  isbn         = {9781467388511},
  language     = {eng},
  pages        = {2894--2902},
  publisher    = {IEEE Computer Society},
  title        = {Reinforcement learning for visual object detection},
  url          = {http://dx.doi.org/10.1109/CVPR.2016.316},
  volume       = {2016-January},
  year         = {2016},
}