Presentation 2007-03-16
Feature selection for object recognition by Histograms of Oriented Gradients
Takuya KOBAYASHI, Akinori HIDAKA, Takio KURITA,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) Histograms of Oriented Gradients (HOG) is one of the well-known features for object recognition. HOG is calculated by taking orientation histograms of edge intensity in a local region. N.Dalal et al. proposed an object detection algorithm in which HOG features were extracted from all locations in an image grid. Linear Support Vector Machine (SVM) was used to design a classifier. In this paper, a subset of HOG features was selected by stepwise forward selection to improve the generalization. The improvement of the recognition rate was confirmed through the experiments using MIT pedestrian dataset.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Object Recognition / Feature Selection / Edge Orientation Histogram / Principal Components Analaysis
Paper # PRMU2006-275
Date of Issue

Conference Information
Committee PRMU
Conference Date 2007/3/9(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 Pattern Recognition and Media Understanding (PRMU)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Feature selection for object recognition by Histograms of Oriented Gradients
Sub Title (in English)
Keyword(1) Object Recognition
Keyword(2) Feature Selection
Keyword(3) Edge Orientation Histogram
Keyword(4) Principal Components Analaysis
1st Author's Name Takuya KOBAYASHI
1st Author's Affiliation Univercity of Tsukuba()
2nd Author's Name Akinori HIDAKA
2nd Author's Affiliation Univercity of Tsukuba
3rd Author's Name Takio KURITA
3rd Author's Affiliation National Institute of Advanced Industrial Science of Technology
Date 2007-03-16
Paper # PRMU2006-275
Volume (vol) vol.106
Number (no) 606
Page pp.pp.-
#Pages 6
Date of Issue