Presentation 2010-12-03
How should we combine detection and tracking to achieve accurate pedestrian recognition?
Hiroki SUGANO, Ryusuke MIYAMOTO,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) Accurate pedestrian recognition is required in collision avoidance of automotive applications and intruder detection systems. Generally, pedestrian recognition is performed by combining detection and tracking. The combination of different two procedures makes it possible to recognize occluded pedestrians and multiple pedestrians. In this paper, we have proposed a novel pedestrian recognition scheme combining CoHOG-based detection and HSV-based tracking using the particle filter. In the proposed scheme, dense proposal distribution required for tracking based on the particle filter is generated by convoluting likelihood obtained from CoHOG-based detection with Gaussian kernel.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Pedestrian Recognition / CoHOG / Particle Filter
Paper # SIS2010-50
Date of Issue

Conference Information
Committee SIS
Conference Date 2010/11/25(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Smart Info-Media Systems (SIS)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) How should we combine detection and tracking to achieve accurate pedestrian recognition?
Sub Title (in English)
Keyword(1) Pedestrian Recognition
Keyword(2) CoHOG
Keyword(3) Particle Filter
1st Author's Name Hiroki SUGANO
1st Author's Affiliation Dept. of Information Systems, Graduate School of Information Science, Nara Institute of Science and Technology()
2nd Author's Name Ryusuke MIYAMOTO
2nd Author's Affiliation Dept. of Information Systems, Graduate School of Information Science, Nara Institute of Science and Technology
Date 2010-12-03
Paper # SIS2010-50
Volume (vol) vol.110
Number (no) 322
Page pp.pp.-
#Pages 6
Date of Issue