Estimation of Pedestrian Pose and Orientation Using on-Board Camera with Histograms of Oriented Gradients Features

Understanding pedestrian behavior, including head and body orientation, is important for a pedestrian safety system. In this paper, we propose an approach that estimates head pose and body orientation by considering two constraints, the pedestrian model constraint between head and body directions and the temporal constraint. In our approach, given an image of pedestrian, image features are extracted and estimates are made of the probabilities of the head position, size and orientation, and the body orientation; these are obtained using a multi-class classifier and tracked by particle filter. We applied two constraints to the particle filter to achieve more accurate estimate. Experiments using real videos from an on-board monocular camera show the effectiveness of our approach.

  • Record URL:
  • Availability:
  • Supplemental Notes:
    • Copyright © 2016, Springer. The contents of this paper reflect the views of the author[s] and do not necessarily reflect the official views or policies of the Transportation Research Board or the National Academy of Sciences.
  • Authors:
    • Yano, Shinya
    • Gu, Yanlei
    • Kamijo, Shunsuke
  • Publication Date: 2016-5

Language

  • English

Media Info

Subject/Index Terms

Filing Info

  • Accession Number: 01598356
  • Record Type: Publication
  • Files: TRIS
  • Created Date: Apr 11 2016 10:21AM