@inproceedings{xiaodongyu-07,
    author = { {X}iaodong {Y}u and {Y}i {L}i and {C}ornelia {F}ermuller and {D}avid {D}oermann },
    abstract = { {T}his paper presents a method for detecting categories of objects in real-world images. {G}iven training images of an object category, our goal is to recognize and localize instances of those objects in a candidate image.
	
	{T}he main contribution of this work is a novel structure of the shape codebook for object detection. {A} shape codebook entry consists of two components: a shape codeword and a group of associated vectors that specify the object centroids. {L}ike their counterpart in language, the shape codewords are simple and generic such that they can be easily extracted from most object categories. {T}he associated vectors store the geometrical relationships between the shape codewords, which specify the characteristics of a particular object category. {T}hus they can be considered as the ``grammar'' of the shape codebook. 
	
	{I}n this paper, we use {T}riple-{A}djacent-{S}egments ({TAS}) extracted from image edges as the shape codewords. {O}bject detection is performed in a probabilistic voting framework. {E}xperimental results on public datasets show performance similar to the state-of-the-art, yet our method has significantly lower complexity and requires considerably less supervision in the training ({W}e only need bounding boxes for a few training samples, do not need figure/ground segmentation and do not need a validation dataset).
 },
    booktitle = { {B}ritish {M}achine {V}ision {C}onference ({BMVC}'07) },
    month = { {D}ecember },
    pdffile = { http://lampsrv02.umiacs.umd.edu/pubs/Papers/xiaodongyu-07/xiaodongyu-07.pdf },
    publisher = { {BMVC}07 },
    title = { {O}bject {D}etection {U}sing {S}hape {C}odebook },
    year = { 2007 }
}