سال انتشار: ۱۳۹۰

محل انتشار: هفتمین کنفرانس ماشین بینایی و پردازش تصویر

تعداد صفحات: ۵

نویسنده(ها):

Maedeh Ahmadi – Artificial Intelligence Laboratory, Dept. of Electrical and Computer Engineering
Maziar Palhang – Artificial Intelligence Laboratory, Dept. of Electrical and Computer Engineering
Niloofar Gheissari – Dept. of Electrical and Computer EngineeringIsfahan University of TechnologyIsfahan, Iran

چکیده:

In this paper, a contour-based object detectionmethod based on Max-Margin Hough transform is proposed. Welearn Implicit Shape Model using local contour features namelyPair of Adjacent Segments (PAS) features. A Max-Margin Houghtransform (M2HT) [1] is then applied, where local parts generateweighted votes for possible object locations. Weights are learnt sothat higher weights are assigned to parts which repeatedlyappear in consistent locations. The achieved results on TUD cowsreference dataset show that discriminative learning of weightsimproves the contour-based Hough detector.